Cell culture
We cultured mESCs in t2iL medium containing Dulbeccos modified eagle medium (DMEM, Nacalai Tesque), 2mM Glutamax (Nacalai Tesque), 1 non-essential amino acids (Nacalai Tesque), 1mM sodium pyruvate (Nacalai Tesque), 100Uml1 penicillin, 100gml1 streptomycin (P/S) (Nacalai Tesque), 0.1mM 2-mercaptoethanol (Sigma) and 15% fetal bovine serum (FBS) (Gibco), supplemented with 0.2M PD0325901 (Sigma), 3M CHIR99021 (Cayman) and 1,000Uml1 recombinant mouse leukaemia inhibitory factor (Millipore)54. A higher PD0325901 concentration of 1M was used for the 2iL medium. mESC colonies were dissociated with trypsin (Nacalai Tesque) and plated on gelatin-coated dishes. Y-27632 (10M, Sigma) was added when cells were passaged. hiPSCs were cultured in mTeSR Plus medium (Veritas). hiPSC colonies were dissociated with Accutase (Nacalai Tesque) and plated on Matrigel-coated dishes (Corning, 3/250 dilution with DMEM). Y-27632 and 1% FBS were added when cells were passaged. WT hiPSCs (409B2, HPS0076) were provided by the RIKEN BioResource Research Centre (BRC)55. FOP hiPSCs (HPS0376) were provided by RIKEN BRC through the National BioResource Project of the Japan Ministry of Education, Culture, Sports, Science and Technology (MEXT) and the Agency for Medical Research and Development (AMED)43. Experiments using hiPSCs were approved by the Kyushu University Institutional Review Board for Human Genome/Gene Research. HEK293T cells and mouse embryonic fibroblasts were cultured in 10% FBS medium containing DMEM, 2mM l-glutamine (Nacalai Tesque), 100Uml1 penicillin, 100gml1 streptomycin (P/S) (Nacalai Tesque) and 10% FBS. hADSCs (Thermo Fisher) were cultured in MesenPRO RS medium (Thermo Fisher). Culture conditions of a HB-AIMS cell line are described in the Generation of AIMS cell lines and mice and AIMS analysis section. Cells were maintained at 37C and 5% CO2.
In this study, we used C57BL/6 mice (Clea Japan), ICR mice (Clea Japan) and R26RYFP/YFP mice (a gift from Frank Costantini at Columbia University, NY, USA)56. The experiments were approved by the Kyushu University Animal Experiment Committee, and the care and use of the animals were in accordance with institutional guidelines.
All primers, spacer linkers and ssODNs used in the present study are listed in Supplementary Table 3.
Mouse ES B6-5-2 and B6-D2-4 cell lines were established from E3.5 blastocysts of the C57BL/6 strain using 2iL and t2iL medium, respectively; an R26RYFP/+ mESC line was established using t2iL medium. Blastocysts were placed on feeders (mitomycin C-treated mouse embryonic fibroblasts) after removal of the zona pellucida. Inner cell mass outgrowths (passage number 0, p0) were dissociated with trypsin and plated on gelatin-coated plates (p1). After domed colonies formed, they were dissociated and passaged (p2). mESC lines were generated by repeating this procedure.
Knock-in (KI) template plasmids for Cdh1-AIMS were generated by attaching the 5 and 3 arms to plasmids containing P2A1:Venus or P2A1:tdTomato cassettes. P2A1 is identical to a widely used P2A sequence26. The 5 arm was designed such that the coding end was fused in-frame to the P2A sequence to allow independent production of both E-cadherin (CDH1) and fluorescence protein. KI plasmids for Tbx3-AIMS were constructed using the same strategy. The alternative P2A sequence P2A2 was constructed by introducing silent mutations to each codon of the original P2A sequence. The conventional CRISPR-Cas9 system was used to efficiently knock-in the dual-colour plasmids in a pair of alleles. A spacer linker was designed to induce a DSB downstream of the stop codon, then inserted into the BpiI sites of a pSpCas9(BB)-2A-Puro (PX459) V2.0 plasmid (Addgene, 62988; see the Plasmid construction section)57. All sgRNAs used in this study were designed using the CRISPR DESIGN (http://crispr.mit.edu/) or CRISPOR tool (http://crispor.tefor.net).
The constructed all-in-one CRISPR plasmids and dual-coloured KI plasmids were co-transfected into mESCs using Lipofectamine 3000 (Thermo Fisher). Dissociated mESCs were plated on gelatin-coated 24-well plates with 500l of (t)2iL+Y-27632 medium ((t)2iL+Y). Nucleic acidLipofectamine 3000 complexes were prepared in accordance with the standard Lipofectamine 3000 protocol. We added 1l of Lipofectamine 3000 reagent to 25l Opti-MEM medium; simultaneously, 250ng of each plasmid (all-in-one, Cdh1-P2A-tdTomato and Cdh1-P2A-Venus plasmid) plus 1l of P3000 reagent were mixed with 25l of Opti-MEM medium in a different tube. These mixtures were combined and incubated for 5min at room temperature, then added to the 24-well plate immediately after cells were seeded. At 24h after transfection, puromycin (1.5 or 2gml1) was added for 2d and then washed out. The transiently treated puromycin-resistant cells were cultured for several days; dual-colour-positive colonies were picked and passaged. Genotypes for the candidate dual KI clones were confirmed by PCR. In this study, transfection experiments for mouse and human cells were performed using this procedure, with passage steps added for an AIMS assay to avoid mosaicism (Fig. 1d). Fluorescence microscopes (BZ-X800 (Keyence) and IX73 (Olympus)) were used to analyse the AIMS data. To extract genomic DNA for clonal sequence analysis, single mESC and hiPSC colonies were suspended in 510l 50mM NaOH (Nacalai Tesque) and incubated at 99C for 10min. PCR was performed using the template genomic DNA, and the amplicons were sequenced by Sanger sequencing.
For generation of AIMS mice, the established dual KI mESC clone (Cdh1-P2A1-tdTomato/Venus AIMS) was dissociated with trypsin and 58 cells were injected into 8-cell embryos (E2.5) collected from pregnant ICR mice. Injected blastocysts were transferred into the uteri of pseudo-pregnant ICR mice and chimaeras were generated. Male chimaeras were mated with C57BL/6 females, and Cdh1-P2A1-tdTomato and Cdh1-P2A1-Venus KI mouse lines were obtained through germline transmission. After the two genotype mice were mated, homozygous AIMS mice were generated.
HB-AIMS cells were established from the E12.5 dual KI embryos according to the protocol of a previous work58 with some modifications. Briefly, the whole liver was mechanically dissociated and filtrated, and the dissociated cells were seeded onto a type I collagen-coated plate (Iwaki) with the HB medium. The HB medium is composed of a 1:1 mixture of DMEM and F-12 (Nacalai Tesque), supplemented with 10% FBS (Gibco), 1gml1 insulin (Wako), 0.1M dexamethasone (Sigma-Aldrich), 10mM nicotinamide (Sigma-Aldrich), 2mM l-glutamine (Nacalai Tesque), 50M -mercaptoethanol (Nacalai Tesque), 20ngml1 recombinant human hepatocyte growth factor (rhHGF) (PeproTech), 50ngml1 recombinant human epidermal growth factor (rhHGF) (Sigma), penicillin/streptomycin (Nacalai Tesque), and small molecules of 10M Y-27632 (Wako), 0.5M A8301 (Tocris) and 3M CHIR99021 (Tocris). After expansion of HBs, a single-cell-derived HB colony with homogeneous expression of tdTomato and Venus was picked and established as an HB-AIMS cell line.
To generate all-in-one CRISPR plasmids for [5C](3A), [10C](8A), [15C](13A), [20C](18C), [25C](23A) and [30C](28A)sgRNA expression, spacer linkers were inserted into the BpiI sites of a PX459 plasmid (Extended Data Fig. 2b). In the plasmids, the 3rd, 8th, 13th, 18th, 23rd or 28th cytosine was replaced with adenine because the overhang sequence of CACC is required for linker ligation. The standard spacer linkers (20nt) or longer spacer linkers (30nt or 40nt) were inserted into the BpiI sites of the [0C], [5C](3A), [10C](8A), [15C](13A), [20C](18A), [25C](23A) or [30C](28A) PX459 plasmid, leading to generation of [5C][30C]sgRNA-expressing all-in-one Cas9 plasmids applicable for puromycin selection. The same [C] linkers were also inserted into the BpiI sites of a PX458 plasmid (Addgene, 62988)57 for selection of GFP-positive transfected cells.
For the plasmid dilution assay, sgRNA-expressing plasmid was constructed by removing a Cas9-T2A-Puro cassette from a PX459 plasmid using the KpnI and NotI sites. Different amounts of sgRNA-expressing plasmid (0250ng) were co-transfected with an unmodified PX459 plasmid (250ng). In addition, [5C][30C] linkers including BpiI sites were inserted into this sgRNA-expressing plasmid to construct [5C][30C]sgRNA-expressing plasmids, which were used for the experiments of CRISPRa (Extended Data Fig. 4e) described below.
For the CRISPR inhibition experiments, the pCMVAcrIIA4 plasmid was generated from the anti-Cas9 AcrIIA4-expressing pCMV+AcrIIA4 plasmid, pCMV-T7-AcrIIA4-NLS(SV40) (KAC200) (Addgene, plasmid 133801)59, by truncating the AcrIIA4 cassette using the NotI and AgeI sites.
For the CRISPRi experiments, the [5C][30C] linkers including BsmBI sites were inserted into the BsmBI sites of an LV hU6-sgRNA hUbC-dCas9-KRAB-T2a-Puro (sgRNA-KRAB-Puro) plasmid (Addgene, 71236)60 to construct [C]sgRNA-expressing all-in-one CRISPRi plasmids. The sgRNA spacers targeting BRCA1 and CXCR4 used in previous studies61 were inserted into the BsmBI sites of the all-in-one plasmids. A puromycin-selectable all-in-one plasmid for CRISPRa was constructed by replacing a GFP cassette of a pLV hU6-gRNA(anti-sense) hUbC-VP64-dCas9-VP64-T2A-GFP (sgRNA-VP64-GFP) plasmid (Addgene, 66707) with a puromycin N-acetyl transferase (PuroR) cassette. A synthetic gene encoding VP64-T2A-PuroR (AZENTA) (Supplementary Table 3) was inserted into the sgRNA-KRAB-GFP plasmid using NheI and AgeI sites, resulting in an sgRNA-VP64-Puro plasmid. In Fig. 4e, the [1C][10C] spacer linkers for targeting ASCL162 were inserted into the sgRNA-VP64-Puro plasmid. In Extended Data Fig. 4e, spacer linkers for targeting ASCL1 and TTN62 were inserted into the BpiI sites of the [0C][30]sgRNA-expressing plasmids, and then they were co-transfected with the spacerless all-in-one CRISPRa plasmid.
To construct all-in-one AsCpf1 plasmids enabling puromycin selection, a synthetic DNA fragment encoding U6 promoter and two BpiI sites (AZENTA) (Supplementary Table 3) was inserted into a PX459 plasmid while removing a U6-gRNA cassette using PciI and XbaI sites. Next, a CBh-Cas9 region of the crRNA-Cas9-puro plasmid was replaced with a CBh-AsCpf1 fragment digested from a pY036_ATP1A1_G3_Array plasmid (Addgene, 86619)63 using KpnI and FseI, resulting in the construction of an all-in-one crRNA-AsCpf1-puro plasmid (PX459 plasmid backbone). The crRNA linkers (Supplementary Table 3) targeting P2A2 sites of AIMS are composed of 5 hairpin, 20nt-spacer and U4AU4 3-overhang, which is known to increase editing efficiency of AsCpf1 (ref. 64), and they were inserted into the BpiI sites of the crRNA-AsCpf1-puro plasmid.
pSpCas9(BB)-2A-Puro (PX459) V2.0 (Addgene, plasmid 62988; http://n2t.net/addgene:62988; RRID: Addgene_62988) and pSpCas9(BB)-2A-GFP (PX458) (Addgene, plasmid 48138; http://n2t.net/addgene:48138; RRID: Addgene_48138) were gifts from Feng Zhang. The pY036_ATP1A1_G3_Array was a gift from Yannick Doyon (Addgene, plasmid 86619; http://n2t.net/addgene:86619; RRID: Addgene_86619). pLV hU6-sgRNA hUbC-dCas9-KRAB-T2a-Puro was a gift from Charles Gersbach (Addgene, plasmid 71236; http://n2t.net/addgene:71236; RRID: Addgene_71236). pLV hU6-gRNA(anti-sense) hUbC-VP64-dCas9-VP64-T2A-GFP was a gift from Charles Gersbach (Addgene, plasmid 66707; http://n2t.net/addgene:66707; RRID: Addgene_66707). pCMV-T7-AcrIIA4-NLS(SV40) (KAC200) was gifted by Joseph Bondy-Denomy and Benjamin Kleinstiver (Addgene, plasmid 133801; http://n2t.net/addgene:133801; RRID: Addgene_133801)59.
To detect sgRNAs complexed with Cas9, 1l of Cas9 (1M) (Alt-R S.p. Cas9 Nuclease V3, IDT) and 1l of synthetic sgRNAs (3M, 1M or 0.3M; IDT) were mixed with 8l of distilled water (total reaction volume of 10l) and reacted on ice for 30min. Samples were loaded onto Bullet PAGE One Precast gels (6%) (Nacalai Tesque) in Tris-borate-ethylenediaminetetraacetic acid (Tris-Borate-EDTA) buffer. RNA was transferred to a Hybond N+ membrane (GE Healthcare) and cross-linked using CX-2000 (Analytik Jena). An sgRNA tracer probe was labelled with an alkali-labile digoxigenin (DIG)-11-deoxyuridine triphosphate (dUTP) using a PCR DIG Probe Synthesis kit (Roche); DNA fragments were amplified using PCR and primers (Supplementary Table 3). After hybridization, specific bands were visualized with the CDP-Star reagent (Roche) using a luminescent image analyser (LAS-3000, FUJIFILM).
To detect DNA fragments complexed with sgRNA-dCas9, we mixed 1l of dCas9 (1M) (Alt-R S.p. dCas9 Nuclease V3, IDT) and 1l of synthetic sgRNAs (1M; IDT) with distilled water for a final reaction volume of 10l, then reacted the mixture at room temperature for 10min. After the reaction, the RNP complex was mixed with 100ng of DNA fragment and 1l of 10 Cas9 reaction buffer (1M HEPES, 3M NaCl, 1M MgCl2 and 250mM EDTA (pH 6.5)), then reacted at room temperature for 10min. The resulting 10l samples were loaded onto 2% agarose gels in Tris-acetate-EDTA buffer; DNA bands were detected by staining with ethidium bromide. The target DNA fragment (647bp) was prepared by PCR amplification from a Tbx3-P2A1-Venus KI plasmid using primers (Supplementary Table 3).
The sgRNA-Cas9-DNA complex was formed using most of the gel shift assay procedure, although its formation also included Cas9 and 3M of synthetic sgRNA. The samples were reacted at 37C for 90min, denatured at 70C for 10min and loaded onto Bullet PAGE One Precast gels (6%) (Nacalai Tesque).
A 20l sgRNA-Cas9-DNA complex was prepared via the procedure used in the gel shift assay. A cleavage reaction was performed at 37C for 30min; a 10l volume was kept on ice while the other 10l volume was denatured at 70C for 10min. The products were loaded onto 2% agarose gels.
Total RNAs were extracted from mESCs at 68h after transfection with P2A1-[C]sgRNA1-PX459 plasmids. Transfected cells were selected by 2d of treatment with puromycin (1.5gml1), then resuspended with ISOGEN II (NIPPON GENE). The samples were incubated for 10min at room temperature, then heated at 55C for 10min. Total RNA was isolated following the manufacturers protocol. After reaction at 70C for 10min, 30g RNAs were loaded onto Extra PAGE One Precast gels (520%) (Nacalai Tesque) in Tris-borate-EDTA buffer. RNA transfer, DIG-probe hybridization and signal detection were performed following the procedure used in the gel shift assay. The DIG probe was labelled by PCR amplification of the DNA fragment (primers shown in Supplementary Table 3). The mU6 DIG-probe was prepared by amplifying the DNA fragment from mESC complementary DNA using specific primers (Supplementary Table 3). cDNA was synthesized using a specific primer that targeted U6 small nuclear RNA65.
Template DNA fragments required for IVT were amplified from a P2A1-gRNA1-PX459 plasmid by PCR (primers shown in Supplementary Table 3). The T7 promoter sequence and cytosine tails were added to the 5-end of the forward primer. We synthesized [0C], [10C] and [25C]sgRNAs using the T7 RiboMAX Express large-scale RNA production system (Promega) following the manufacturers protocol.
FIJI software was used to quantify band signals for the gel shift, DNA cleavage and northern blot assays.
The PX458-based all-in-one plasmids (250ng) for targeting VEGFA1 gene were transfected into hADSCs using Lipofectamine 3000 upon 80% confluency. Immediately after adding the plasmid:Lipofectamine mixture into the cells, the plates were centrifuged at 700g at 35C for 10min to increase transfection efficiency. The cells were cultured for 7d without passaging to allow continuous expression of the plasmid, and then GFP-positive single cells were picked using a hand-made capillary and transferred to PCR tubes (1 cell per tube). To enable sequence analysis for a pair of alleles from a single cell, whole genomic DNA were amplified using PicoPLEX (TAKARA) according to the manufacturers instructions. The genomic locus targeted by Cas9 was amplified by PCR using primers (Supplementary Table 3) and the PCR amplicons were sequenced.
At 24h after transfection with the all-in-one Cas9 plasmid, mESCs were treated with the Cas9 inhibitor BRD0539 (TOCRIS) during puromycin selection and subsequent culture until analysis.
pCMV+AcrIIA4 plasmid was co-transfected with 250ng of the all-in-one Cas9 plasmid in different amounts (2.52,500ng for 24-well plates). For the BRD0539 and AcrIIA4 experiments, puromycin selection and indel analysis were performed using the same procedure as described above (Generation of AIMS cell lines and mice and AIMS analysis section) and in Fig. 1d.
A day before transfection, 3104 HEK293T cells were seeded onto a 96-well plate. The all-in-one CRISPRa/i plasmids (50ng, 1/5 scale of the 24-well plate version) were transfected and cultured for 24h. Then, puromycin (5.0gml1) was treated for 2d to exclude untransfected cells. After removal of puromycin, the transfected cells were cultured for 1d and 2d for CRISPRa and CRISPRi, respectively, and total RNAs were extracted using ISOGEN II as described above (Northern blotting section).
The cDNAs were synthesized from total RNAs using SuperScript III Reverse Transcriptase (Thermo Fisher) according to the manufacturers instructions. RTqPCR was conducted using a THUNDERBIRD SYBR qPCR Mix (Toyobo) and CFX Connect real-time PCR detection system (BIO RAD) according to the manufacturers instructions. Primers for ASCL1, TTN, BRCA1 and CXCR4 used in previous studies61,62, and for GAPDH are listed in Supplementary Table 3. The values for GAPDH were used as normalization controls.
A Tbx3-P2A1-tdTomato KI plasmid was co-transfected with Tbx3-sgRNA1-expressing PX459 to the mESCs. After transient puromycin selection, colonies were dissociated and passaged; the resulting colonies were analysed. Colonies with mosaic tdTomato expression were excluded from data analysis. After the colonies had been counted, positive tdTomato colonies were selected and genomic DNA was extracted for sequencing.
The neomycin (Neo) KI plasmid was constructed by replacing the tdTomato cassette of the Tbx3-P2A1-tdTomato KI plasmid with a P2A1-Neo cassette. The KI plasmid was co-transfected with P2A1 sgRNA1-expressing PX459 to a Tbx3-P2A1-AIMS clone. When puromycin was removed, geneticin (400gml1, Gibco) was added to select KI clones. All eight clones were confirmed to possess KI genotypes; geneticin-resistant colonies were identified as KI.
PCR reactions to amplify specific on-target or off-target sites were performed using KOD-Plus-ver.2 DNA polymerase (Toyobo) in accordance with the manufacturers protocol. The resulting PCR amplicons were denatured and re-annealed in 1 NEB buffer 2 (NEB) in a total volume of 9l under the following conditions: 95C for 5min, reduction from 95C to 25C at a rate of 0.1Cs1 and indefinite incubation at 4C. After re-annealing had been performed, 1l of T7 endonuclease I (NEB, 10Ul1) was added and the product was incubated at 37C for 15min.
Purified PCR products to amplify specific on-target or off-target sites were inserted into a T-easy vector (Promega) and transformed into DH5- bacterial cells. For rapid and efficient indel detection, plasmids were directly isolated from each white colony after blue/white screening; the inserted DNA fragment was amplified by PCR. The PCR amplicons were mixed with PCR products amplified from a WT DNA template such as KI plasmid or unedited genomic DNA; a T7E1 assay was then performed. Sanger sequencing was also performed for PCR amplicons that were not digested by T7E1 to determine the total number of colonies that harbour indels. The Bac[P] value was calculated as follows: Bac[P]=Indel/Total.
Bac[P] values for both WT and R206H alleles were determined through indel induction experiments using various [C]sgRNAs in the mESC clone of the FOP model. The targeting sites of both WT and R206H alleles were amplified by PCR, then cloned into a T-easy vector. Sanger sequencing was performed for each PCR product that had been derived from single bacterial clones, as described above. Similarly, Bac[P] values for both R206H (pf) and WT (1mm) alleles were determined by inducing indels in FOP hiPSCs; a corrected cell line (WT/Corrected) was used to determine the Bac[P] value of the corrected allele (2mm). Some PCR products did not contain a G/A hallmark because of intermediate-sized deletions (12~50 nucleotides); it was therefore impossible to determine which allele was edited for these PCR products. We observed that the fraction of such products with intermediate-sized deletions was generally constant (~20% in experiments shown in Fig. 6 and 1020% in experiments shown in Fig. 7) and did not decrease with [C] extension, suggesting that such intermediate-sized deletions are byproducts of the short indel induction processes. Therefore, we assigned products with intermediate-sized deletions to two alleles using the ratio of PCR products with convincingly confirmed origins. For the analysis shown in Fig. 7, we calculated the means of Bac[P] for WT (1mm) alleles on the basis of comparisons of R206H (pf) to WT (1mm) alleles and WT (1mm) to corrected (2mm) alleles for subsequent computational analyses.
Using the transfection protocol described above (Generation of AIMS cell lines and mice and AIMS analysis), 2105 WT hiPSCs or 4104 HEK293T cells were seeded onto 48-well plates and transfected with 100ng of all-in-one CRISPR plasmids (2/5 scale of the 24-well plate version). hiPSCs were dissociated and counted using trypan blue at 3 or 4d after transient puromycin treatment (1.5gml1); HEK293T cells were counted at 4d after transient puromycin treatment (3gml1). The data obtained by this procedure are indicated as Cell number in the Figures.
Biochemical assays were also performed using Cell Count Reagent SF reagent according to the manufacturers instructions (Nacalai Tesque). The Cdh1-P2A1-AIMS mESCs (2104 cells) were seeded onto 96-well plates and transfected with 50ng of all-in-one plasmids (1/5 scale of the 24-well plate version). Two days after puromycin selection, absorbance at 450nm was measured by Multiskan FC (Thermo Fisher). The data obtained from the biochemical assay are indicated as Cell viability (%) in Fig. 4d and Extended Data Fig. 4c by setting the data for [0C] and 0mM as a reference value (1.0), respectively.
For the AcrIIA4 experiments (Fig. 4c and Extended Data Fig. 4b), the Cdh1-P2A1-AIMS mESCs (3104 cells) were seeded onto 96-well plates and 50ng of all-in-one plasmids were co-transfected with different amounts of pCMV+AcrIIA4 and/or pCMVAcrIIA4 plasmids (1/5 scale of the 24-well plate version). In Fig. 4c and Extended Data Fig. 4b, we observed cytotoxicity for higher doses of AcrIIA4 expression plasmids. Similar cytotoxicity profiles were obtained in the absence of the Cdh1-P2A1-sgRNA1 target sequence in WT mESCs.
The transfection protocol for the 24-well plate experiment was performed as described above (Generation of AIMS cell lines and mice and AIMS analysis). For HDR induction in mESCs, WT hiPSCs and HEK293T cells, 1l of 10M ssODN (Eurofins) was added to the plasmidLipofectamine complex; for hiPSC transfection, 1l of 3M ssODN was added because a concentration of 10M induced severe toxicity. After transient puromycin selection, colonies were dissociated and plated at low density to avoid mosaicism. Single colonies were selected and genomic DNA was extracted. Sequence analysis was performed to identify G to A replacement with or without indels. To correct the FOP hiPSCs, clones that underwent HDR were screened by digesting the PCR product using the BstUI restriction enzyme (NEB); BstUI-positive PCR products were then sequenced. A silent mutation was inserted into the ssODN to generate the BstUI site and to distinguish an HDR-corrected (Corrected) allele from an original WT allele. Without this hallmark, WT/ clones, in which PCR amplicons from the R206H allele cannot to be obtained because of large deletions or more complex genomic rearrangement, would be misidentified as WT/Corrected clones.
For p53 staining, we performed transfection for HDR induction (1/5 scale of the 24-well plate version), using the protocol described above. In this assay, 6104 hiPSCs were seeded on a Matrigel-coated 96-well plate in triplicate. Puromycin selection was performed to examine p53 activity solely in transfected cells. The surviving cells were fixed with 4% paraformaldehyde at 2d after puromycin removal. For pSmad1/5/8 staining, 5103 cells were plated on a Matrigel-coated 96-well plate without Y-27632 and with 1% FBS. After 2.5h of culture, activin-A (100ngml1) (R & D Systems) was administered for 30min; cells were fixed with 4% paraformaldehyde. Antibody reactions were performed in accordance with standard protocols. Rabbit polyclonal p53 (FL-393, Santa Cruz, 1:200) and rabbit monoclonal pSmad1/5/8 (D5B10, Cell Signaling Technology, 1:1,000) antibodies were reacted overnight at 4C. Donkey anti-rabbit Alexa Fluor 488 secondary antibody (Thermo Fisher, 1:1,000) was reacted at room temperature for 30min. Data analysis was performed using a cell count application associated with a fluorescent microscope to select cells with p53 and pSmad1/5/8 activation by means of fluorescence intensity thresholds (BZ-X800, Keyence).
An mESC clone of an FOP model (R26RYFP/+ mESC line) was dissociated with trypsin and 58 cells were injected into 8-cell embryos (E2.5) collected from pregnant ICR mice. Injected blastocysts were transferred into the uteri of pseudo-pregnant ICR mice. Chimaeric contribution was confirmed by coat colour and YFP fluorescence. YFP was observed using a fluorescence stereo microscope (M165FC, Leica).
In this study, the probability of single-allele editing (P) was determined using AIMS and a Bac[P] assay, on the basis of a T7E1 assay, complemented by sequence validation. AIMS-based P (AIMS[P]) was determined as follows:
$$begin{array}{*{20}{c}} {mathrm{AIMS}left[ {mathrm{P}} right] = frac{{left( {2Fleft( {mathrm{Bi}} right) + Fleft( {mathrm{Mono}} right)} right)}}{2}} end{array}$$
(1)
where F(Bi) and F(Mono) are the experimental frequencies of cells with bi-allelic and mono-allelic genome editing, respectively.
The efficiency of the single-allele editing P (P(pf), where pf denotes perfect match) can be described as follows:
$${{mathrm{P}}left( {pf} right) = frac{S}{{K + S}}}$$
(2)
where the concentration of effective sgRNA-Cas9 complexes and the dissociation constant between the sgRNA and its target site are defined as S and K, respectively. On the basis of high editing efficiency without [C] extension (P=approximately 1), we assumed that the recovery rate from single-site damage was very low; therefore, it was neglected in subsequent analyses. To mechanistically understand the effects of [C] extension and 1mm, we assumed that [C] extension and 1mm decreased S and increased K, respectively. By setting S=1 for each sgRNA sequence without [C] extension, we approximated K values for each of eight sgRNA sequences. When P (AIMS[P] or Bac[P]) was 1, P was set to 0.99. Next, the relative S concentrations were determined using K and AIMS[P] for sgRNAs with [C] extension. Despite variation in the relationships between [C] extension and AIMS[P] among sgRNA sequences (Fig. 2f), we found clear and similar inverse relationships between [C] extension and relative S values for different sgRNA sequences (Extended Data Fig. 3d). Linear regression analysis demonstrated a good fit for the logarithm of the ratio of S to the length of [C] extension for all sgRNA sequences (Fig. 2g). Analysis of covariance (ANCOVA) indicated that the linear regression slopes did not significantly differ among various sgRNA sequences (Fig. 2h). This finding suggests that [C] extension exerts uniform suppression effects on diverse sgRNA sequences.
Since we observed that [C] extension modestly decreased target cleavage (Fig. 3c), we also performed similar analysis by gradually increasing K according to the length of [C] extension and observed that [C] extension gradually decreases S in a similar manner. In this setting, the effects on S became weaker. However, we observed that the dynamic range of suppression in northern blot analysis (Fig. 3f, ~6,000-fold change at [30C]) was more comparable to the range of change in S with constant K (~2,000-fold change at [30C]) relative to the range of change in S with increased K (~400-fold and 200-fold change with 5-fold and 10-fold increases in K at [30C], respectively). Therefore, this suggests that the effects on complex formation may be dominant, allowing determination of the single-allele editing probability in the cells.
In the initial phase of this study, we compared matched AIMS[P] and Bac[P] values for nine sgRNAs (that is, Cdh1-P2A1-sgRNA1 with different [C] extension lengths) and observed that AIMS[P] was strongly correlated with Bac[P] (Extended Data Fig. 5a). In our subsequent analyses, we used AIMS[P] to model indel insertion frequency (Figs. 2 and 5, and Extended Data Fig. 6) and Bac[P] to model HDR frequency (Figs. 6 and 7).
AIMS error was calculated as the difference between raw AIMS[P] and adjusted AIMS[P] (adjusted AIMS[P]AIMS[P]) (Fig. 1h). The raw AIMS[P] is simply based on fluorescence patterns. Therefore, in Fig. 1e, rare tdTomato+/Venusindel and tdTomatoindel/Venus+ heterozygous clones were grouped into mono-allelic clones. To determine the exact number of bi-allelic indel clones, these ostensibly heterozygous clones were analysed for sequencing (Seq-indel data). When sequencing these clones, most (86%) of these ostensibly heterozygous clones turned out to be homozygous. Adjusted AIMS[P] incorporates Seq-indel data together with fluorescence patterns. In most analyses, we used raw AIMS[P].
T7E1 error was calculated as Bac[P]T7E1:Bac[P] (Fig. 1i,j). T7E1:Bac[P] is the indel probability calculated from the rate of T7E1 sensitive clones, while Bac[P] is the indel probability calculated considering the Seq-indel data. The Seq-indel data were the exact numbers of indel clones that were not digested by T7E1, as determined by sequencing PCR products.
We performed extensive analyses using a combination of AIMS and sgRNAs with various types of [C] extensions. When editing efficiency was homogeneous across the cell population, we estimated the frequencies of cells with bi-allelic, mono-allelic or no genome editing (that is, F(Bi), F(Mono) or F(No)) as follows:
$${F(mathrm{Bi}) = mathrm{AIMS}[{mathrm{P}}]^2}$$
(3)
$${F(mathrm{Mono}) = 2mathrm{AIMS}[{mathrm{P}}]left( {1 - mathrm{AIMS}left[ {mathrm{P}} right]} right)}$$
(4)
$${Fleft( {mathrm{No}} right) = left( {1 - mathrm{AIMS}left[{mathrm{P}}right]} right)^2}$$
(5)
Using these equations, we observed that actual F(Mono) was lower than estimated F(Mono), particularly at intermediate AIMS[P] levels (AIMS[P]=~0.5). Therefore, we considered genome editing frequency heterogeneity at the single-cell level, which we modelled using a beta distribution. The probability density functions of P and mean P (E(P)) were calculated as follows:
$${fleft( {{mathrm{P}};alpha ,beta } right) = frac{{{mathrm{P}}^{alpha - 1}left( {1 - {mathrm{P}}} right)^{beta - 1}}}{{Bleft( {alpha ,beta } right)}}}$$
(6)
$${Eleft( {mathrm{P}} right) = frac{alpha }{{alpha + beta }}}$$
(7)
where the mean P corresponds to AIMS[P] (or Bac[P]) and and are exponents of P and its complement to 1. Using the beta distribution, F(Bi), F(Mono) and F(No) were described as follows:
$${F(mathrm{Bi}) = mathop {smallint }limits_0^1 {mathrm{P}}^2fleft( {{mathrm{P}};alpha ,beta } right)dP}$$
(8)
$${F(mathrm{Mono}) = mathop {smallint }limits_0^1 2{mathrm{P}}(1 - {mathrm{P}})fleft( {{mathrm{P}};alpha ,beta } right)dP}$$
(9)
$${Fleft( {mathrm{No}} right) = mathop {smallint }limits_0^1 (1 - {mathrm{P}})^2fleft( {{mathrm{P}};alpha ,beta } right)dP}$$
(10)
Using these equations, we determined values for each experiment that minimized the squared residuals between experimental F(Bi), F(Mono) and F(No), and simulated F(Bi), F(Mono) and F(No) (Extended Data Fig. 5b). As shown in Extended Data Fig. 5b, we observed that optimized values were generally constant for a wide range of AIMS[P] (0.1
As described above, 1mm (or 2mm) increases K in equation (2). The efficiency of the single-gene editing P on the 1mm (or 2mm) target can be described as follows:
$${{mathrm{P}}left( {1mathrm{mm};or;2mathrm{mm}} right) = frac{S}{{mK + S}}}$$
(11)
where m is the ratio of K for the 1mm target to K for the perfect match target. Thus, the single-gene editing P for 1mm (or 2mm) can be expressed as the function of P(pf), as follows:
$${{mathrm{P}}left( {1mathrm{mm};or;2mathrm{mm}} right) = frac{{{mathrm{P}}left( {pf} right)}}{{left( {1 - m} right){mathrm{P}}left( {pf} right) + m}}}$$
(12)
For the results shown in Figs. 6 and 7, we determined values of m that fit P(pf) and P(1mm or 2mm), using SSR as the error function (Fig. 6g). The ratios of P(pf) and P(1mm or 2mm) can also be described as functions of P(pf), as follows:
$${frac{{{mathrm{P}}(1mathrm{mm};or;2mathrm{mm})}}{{{mathrm{P}}(pf)}} = frac{1}{{left( {1 - m} right){mathrm{P}}left( {pf} right) + m}}}$$
(13)
$${frac{{{mathrm{P}}left( {pf} right)}}{{{mathrm{P}}left( {1mathrm{mm};or;2mathrm{mm}} right)}} = left( {1 - m} right){mathrm{P}}left( {pf} right) + m}$$
(14)
As shown in Fig. 6h, decreasing P(pf) contributes to the reduction in relative off-target ratio and enhancement of specificity. Thus, reduction in CRISPR-Cas9 activity through [C] extension is beneficial for reducing the relative off-target activity and enhancing specificity.
Using the beta distribution, the frequencies of the various HDR clones shown in Fig. 6 were determined as follows (Extended Data Fig. 7c,d):
$${F(mathrm{WT}/mathrm{R206H}) = mathop {smallint }limits_0^1 2h{mathrm{P}}(1 - {mathrm{P}})(1 - (1 - h){mathrm{P}}^{prime} )fleft( {{mathrm{P}};alpha ,beta } right)dP}$$
(15)
$${F(mathrm{WT}/mathrm{R206H} + mathrm{indel}) = mathop {smallint }limits_0^1 2h(1 - h){mathrm{P}}(1 - {mathrm{P}}){mathrm{P}}^{prime} fleft( {{mathrm{P}};alpha ,beta } right)dP}$$
(16)
$${F(mathrm{indel}/mathrm{R206H}) = mathop {smallint }limits_0^1 2h(1 - h){mathrm{P}}^2(1 - (1 - h){mathrm{P}}^{prime})fleft( {{mathrm{P}};alpha ,beta } right)dP}$$
(17)
$${F(mathrm{indel}/mathrm{R206H} + mathrm{indel}) = mathop {smallint }limits_0^1 2h(1 - h)^2{mathrm{P}}^2{mathrm{P}}^{prime} fleft( {{mathrm{P}};alpha ,beta } right)dP}$$
(18)
$${F(mathrm{R206H}/mathrm{R206H}) = mathop {smallint }limits_0^1 h^2{mathrm{P}}^2fleft( {{mathrm{P}};alpha ,beta } right)dP}$$
(19)
$${Fleft( {mathrm{overall};mathrm{HDR}} right) = mathop {smallint }limits_0^1 left( { - h^2{mathrm{P}}^2 + 2hP} right)fleft( {{mathrm{P}};alpha ,beta } right)dP}$$
(20)
where the efficiency of HDR on the Cas9-cleaved single allele is defined as h. The probability of single-gene editing on the edited (that is, 1mm) target is P' (Extended Data Fig. 7d), which is described in a manner similar to equation (12), as follows:
$${mathrm{P}^{prime} = frac{{mathrm{P}}}{{left( {1 - m} right){mathrm{P}} + m}}}$$
(21)
where m=1.723. P is decreased according to the [C] extension length (Extended Data Fig. 7e).
For simplicity, we considered h to be constant across the cell population in each experiment. On the basis of the experimental overall HDR frequency results and equation (20), we estimated h for each [C] extension (Fig. 6f). Although h was very low for sgRNAs without [C] extension (2.07%), h for sgRNAs with [C] extension was generally high (~11%). This result suggests that the conventional system without [C] extension suppresses HDR; [C] extension releases this suppression to allow HDR to reach its upper limit. On the basis of these findings, we used the mean estimated h (10.99%) for [C]-extended sgRNAs; we estimated the frequencies of distinct HDR patterns, overall HDR and precise HDR (Fig. 6i,j). For sgRNAs without [C] extension, we used the estimated h (2.07%). The simulated data adequately fit the experimental results (Fig. 6ik). To predict continuous HDR outcomes, we designed a hypothetical function for h for the range of P, such that h=2.07% for P>0.9 and h=10.99% for P<0.9 (Extended Data Fig. 7f); we estimated the frequencies of distinct HDR patterns, overall HDR and precise HDR (Extended Data Fig. 7g). In the simulation, precise HDR reached a maximum at P=0.313 (Extended Data Fig. 7e,g).
Read more:
Optimization of Cas9 activity through the addition of cytosine ... - Nature.com
